[CONTRIBUTION FROM
THE
LEDERLELABORATORIES DIVISION,AMERICAN CYANAMID Co.]
AN ANTIMALARIAL ALKALOID FROM HYDRANGEA. XXI. SYNTHESIS AND STRUCTURE OF FEBRIFUGINE AND ISOFEBRIFUGINE B. R. BAKER, FRANCIS J. McEVOY, ROBERT E . SCHAUB, JOSEPH P. JOSEPH, AND JAMES H. WILLIAMS Received August 4, 196.2
A second synthesis of the dl-alkaloid starting with furfural has been presented in the preceding paper (1). In this communication is described the synthesis of the naturally occurring antipode of the alkaloid (febrifugine) via optically active intermediates. From the filtrate of the final product was also isolated the isomeric isofebrifugine. Structures for these two isomeric compounds are proposed on the basis of infrared data.
~CHCH,COOH
0 1 NHCOC6H5
I From an economic point of view it would be best to resolve at the earliest possible step. N-Benzoyl-P-furyl-&alanine (I) was resolved in alcohol with brucine to give a good yield of the I-form as the insoluble isomer. It should be noted that this molecule has only one asymmetric center, A, (see Chart I)
- l +- 2+CHART I
A+BI
A+B+-
I1
and that the second asymmetric center, B, is introduced by subsequent hydrogenation to 11. Reduction of dl-I gave almost exclusively one of the two possible racemates of I1 (1). Whether this predominant racemate of I1 has configuration 1 or 2 does not matter in the following argument. If it is assumed that racemate 1 is the main reduction product, then reduction of optically active I must also give the optically active half of racemate 1, that is A (+) --f AB (+,+) and A (-) -+AB (-,-). This has been found to be the case. Recrystallization of crude I1 led to the pure optical antipode of 11,[a]:" - 15.5". This pure antipode was converted to the optically active bromolactone (111) in 63% yield. However, a better over-all conversion of I to I11 (43%) was again obtained as in the dl-series (1) when the intermediate I1 was not purified. The synthesis of the optically activn alkaloid (V) was completed in the same 178
AN ANTIMALARIAL ALKALOID FROM HYDRANGEA. XXI
179
manner as described in the dl-series (1). This compound, obtained in 59% yield by 48% hydrobromic acid hydrolysis of the methyl ether (IV), and the natural Hydrangea alkaloid (febrifugine) (2) had identical I.R. spectra, activity, and rotations. The free base of V, m.p. 158-160°, was also identical with one of the crystal forms of the alkaloid free base as were their N-carbamyl derivatives prepared with potassium cyanate (3).
Is
Iz
From the filtrate of V was isolated an isomeric base, m.p. 138-139", [cy]:' +121° (CHC13),in 13% yield. This base is identical with the isofebrifugine of Koepfli, Mead, and Brockman (4) as shown by their I.R. spectra.' With both febrifugine (the Hydrangea alkaloid) and isofebrifugine on hand, it becomes interesting to speculate why these two isomeric compounds are formed in the same reaction mixture and what their respective structures might be. The two important differences in properties between these bases (4) are as follows: (a) Febrifugine will readily form ketone derivatives while isofebrifugine will not under the same conditions. (b) Isofebrifugine is 14 times more easily extractable from water with chloroform than is febrifugine.
PT
m
There are two major possible explanations for the structures of these two easily interconvertible isomers. Koepfli, Brockman, and Moffat ( 5 ) proposed that the structure VI actually exists as the hemi-ketal (VII). This ketal has a third asymmetric center resulting in twbo diastereoisomers. They proposed that febrifugine and isofebrifugine were the two ketal isomers and that the open hydroxy ketone (VI) did not exist. A second explanation, which they did not mention, is that febrifugine has the open hydroxy ketone structure (VI:) and isofebrifugine is one of the two possible ketal isomers (VII). This can explain both properties (a) and (b) 1 We wish t o thank Dr. J. B. Koepfli for hit3 generosityin supplying t o us samples of isofebrifugine and febrifugine (m.p. 138-140') for comparative purposes. The latter had an I.R. spectrum identical with a sample of the free base, m.p. 135-136", prepared from the Hydrangea alkaloid (dihydrochloride) (2).
180
BAKER, McEVOY, SCHAUB, JOSEPH, AND WILLIAMS
cited above as well as the ease of interconversion. These points will be discussed later. In order to substantiate one or the other of these two hypotheses a study of the infrared spectra of the isomeric bases was made. Kuehl, Spencer, and Folkers (6) stated that febrifugine and isofebrifugine had identical I.R. spectra, but reported only a narrow range of the spectrum (6.03-6.78 p ) . That these two compounds would have identical curves over a wide region (2-15 p ) of the spectrum seemed highly improbable in view of their differences in physical and chemical properties. The isomeric bases showed some similarities in their I.R. spectra, particularly in the 5.9-6.7 p region where they were essentially identical as previously cited (6). Both had the tert-amide band of the 4-quinazolone carbonyl at 5.90-5.96 I.( and the combined phenyl and C--hT bands a t 6.20 I.(. However, there were many gross differences. For example, both crystal forms of febrifugine, m.p. 138-140" and 158-160", had a ketone band a t 5.79 p which was missing in isofebrifugine. Secondly, isofebrifugine had strong bands a t 9.05 and 9.48 p not present in febrifugine. These have been assigned to a five membered cyclic ketal ring (9). Thus febrifugine must have the open hydroxy ketone structure (VI) and isofebrifugine is a hemi-ketal (VII). The mere formation of a hemi-ketal would hardly be sufficient to stop isofebrifugine from forming ketone derivatives of febrifugine since in the sugar series carbonyl derivatives are readily formed. However, inspection of models of isofebrifugine showed that the two possible diastereoisomers, VI11 and IX, could be further stabilized by hydrogen bonding of the OH to the carbonyl of the 4-quinazolone, hence decreasing the potentiality of formation of ketone derivatives and increasing the ease of extractability into organic solvents from water. It cannot be stated a t this time whether isofebrifugine has structure VI11 or IX. These structures require that a third isomer, the diastereoisomer of isofebrifugine, should exist. This isomer has not yet been isolated from a natural source (2, 4, 6, S), from interconversions (4, 6), or from a synthetic reaction mixture such as described in the experimental. Secondly, either hemi-ketal, VI11 or IX, should give ketone derivatives of febrifugine under the proper conditions of simultaneous interconversion.
pm
l x
Acknowledgement: The authors wish to thank Dr. Robert Gore of the Stamford Laboratories of this company for consultation on the interpretation of
AN ANTIMALARIAL ALKALOID FROM HYDRANGEA. XXI
181
the infrared spectra. They are also grateful to W. McEwen and J. Poletto for large scale preparation of some of the intermediates, to L. Brancone and staff for the microanalyses, to Dr. R. Hewitt and co-workers for the antimalarial assays, and to Mr. W. Fulmor and staff for the infrared spectra. EXPERIMENTAL
Resolution of N-benzoyl fl-fury1-8-alanine (I). A hot solution of 145 g. of dl-I and 263 g. of anhydrous brucine in 1450 cc. of absolute alcohol was allowed to cool t o room temperature After being seeded, the mixture R-as allowed t o stand overnight. The brucine salt was collected and washed with 145 cc. of absolute alcohol; wt. 164 g., [a]: -46" (1% in acetone). The rotation of a similar preparation was the same and was unchanged after recrystallization from absolute alchol. A hot solution of 71 g. of brucine salt in 100 cc. of absolute alcohol was poured into a mixture of 500 cc. of water and 25 cc. of 12 N hydrochloric acid with stirring. The mixture was cooled t o O", filtered, and the white cryi3tals were washed with water; yield, 26.8 g. (75% over-all), m.p. 191-192'. Another preparation, m.p. 193", gave -44' (1% in acetone) and -66' (0.8% in 0.1 N NaOH). Anal. Calc'd for C14H13N04: C, 64.9;H, 5.02;N,5.41. Found: C , 64.9;H, 5.39;N, 5.41. The filtrate from the brucine salt deposited additional crystals after standing several weeks. One recrystallization from absolute alcohol gave the optically pure salt of the 1-acid and brought the yield of resolution t o near quantitative. 1-8-Benzamido-8-tetrahydrofurylpropionic acid (11). A mixture of 30 g. of 1-1, 150 cc. of Methyl Cellosolve, and 5 g. of 10% palladium-charcoal was shaken with hydrogen a t 2-3 atm. and 80"until reductionwas complete (one hour). The filtered solutionwas evaporated t o dryness in vacuo leaving a crystalline residue in quantitative yield. Material of this purity was used in subsequent steps. Optically pure material of constant m.p. was obtained with considerable loss after five recrystallizations from 30% methanol; white plates, m.p. 182183", [a]E -15.5' (0.8% in 0.1 A' NaOH). Anal. Calc'd for ClhH1rNO4: C, 63.9;H, 6.46;N, 5.32. Found: C, 64.0;H, 6.83;N,5.23. l-S-Amino-4-hydroxy-7-bromohepatanoic acid lactone (111). A solution of 10.9 g. of nearly optically pure LII, m.p. 178-179', in 10.9 cc. of 48% hydrobromic acid was treated as described for dl-I11 (1);yield, 7.9 g. (63%) of white crystals, m.p. 201-202°dec.,[~]~ -36.6' (0.8% in H90). Anal. Calc'd for CI7H12BrNO2.HBr:C, 27.7;H, 4.29;N, 4.62. Found: C, 27.8; H, 4.35;N , 4.75. For preparative purposes crude I1 obtained on evaporation of the hydrogenation solution was treated with 48% hydrobromic acid giving 4143% yields over-all from 1-1. 1-9-Hydroxypiperidine-2-acetic acid lactone hydrochloride. Cyclization of 3.0 g. of 1-111 with triethylamine as described for the dl-isomer (1) gave 0.93 g. (60%) of product, m.p. 234-235" dec., from absolute alcoholic hydrogcm chloride. Recrystallization from methanol afforded white crystals, m.p. 238-239" dec., [ct]; -64.4' (0.7% in H20). A n a l . Calc'd for C,H11N02.HCl: C, 47.3;IT, 6.75;N , 7.89. Found: C, 47.6;H, 6.87;N , 8.13. d-1-Benzoyl-9-hydroxypiperidine-R-acetica d lactone ( A ) .Benzoylation of 500 mg. of the preceding compound as described (1) forthe dl-isomer, procedure C,gave 580 mg. (84%) of product from ethyl acetate-heptane, m.p. 110". Recrystallization from toluene afforded white crystals of unchanged m.p., [a]:' +113" (1% in abs. alc.). Anal. Calc'd for Cl4HlsNOa: C, 68.6;H, 6.12;N,5.71. Found: C,68.5;H,6.36;N, 5.65. (B).By cyclization of 33.8 g. of E-I11 wkh triethylamine in chloroform followed by benzoylation as described for the dl-isomer, procedure D, (1) there was obtained 20.6 g. (75%) of white crystals, m.p. 106-108". I n two more runs the yields were 66% in both cases.
182
BAKER, MCEVOY, SCHAUB, JOSEPH, AND WILLIAMS
d (8) -1-BenzoyZ-d-hydroxypiperidine-2-acetic acid. Hydrolysis of the preceding lactone with 10% sodium hydroxide as described for the dZ-isomer (1) gave an 82% yield of product, map.156157" dec. This material could not be further purified. I n some runs the product did not crystallize in which case the oil was extracted with chloroform and carried to the next step. d-1-Benzoyl-3-methoxypiperidine-d-acetic acid. By methylation of 10 g. of d-l-benzoyl-3hydroxypiperidine-2-acetic acid as the disodium salt as previously described for the dlisomer (1) there was obtained 4.8 g. (46%) of product, m.p. 149-150", and 2.7 g. (29%) of lactone, m.p. 103-105", was recovered. Recrystallization of a sample from ethyl acetate did not raise the m.p.: white crystals, [a]," 51' (1% in 0.1 N NaOH). Anal. Calc'd for CISH1VNOI: C, 65.0; H, 6.91; N, 5.05. Found: C, 64.6; H, 7.19; N, 5.28. d (?)-1-Carbeth~xy-S-methoxypiperidine-2-acetic acid. By hydrolysis and carbethoxylation of 13.4 g. of the preceding benzoyl acid as described for the dl-isomer (1) there was obtained 9.6 g. (82%) of product as a gum. Anal. Calc'd for C11HloNOa: C, 53.9; H, 7.75; N, 5.71. Found: C, 54.0;H, 7.96; N , 5.62. After standing for some time the gum solidified. Recrystallization from benzene-petroleum ether gave a 72% recovery of white crystals, m.p. 79-80'. d (?)-l-Carbethoxy-l-(~-bromoacetonyl)-3-methoxypiperidine. From 7.0 g. of l-carbethoxy3-methoxypiperidine-2-acetic acid (m.p. 79-80') via the acid chloride and diazoketone as described for the dl-isomer (7) there was obtained 9.0 g. (98%) of product as an oil. d ( a ) -S-w-Keto-(l -carbethoxy-S-methoxy-2-piperidyl)propyl]-~-quinazoZone. Condensation of 9.0 g. of the above bromo ketone with 4-quinazolone a8 described for the dl-isomer (7) gave 5.1 g. (61%) of product, m.p. 121-122". Recrystallization of a similar preparation from benzene-heptane afforded white crystals, m.p. 124-125'. Anal. Calc'd for C ~ O H Z ~ C, ? ~ 62.0; O ~ :H, 6.46; N , 10.9. Found: C, 62.1; H, 6.62; N , 10.8. d (I)-3-~-Keto-~-(S-methoxy-2-piperidyl)propyll-~-quinazolone dihydrochloride (IV). Hydrolysis of 5.1 g. of the preceding compound as described for the dl-isomer (7) gave 1.8 g. (36%) of white crystals which partially melt a t 140' and decompose at 210-212'. A similar preparation was analyzed. Anal. Calc'd for C U H , I N ~ O ~ . ~ H C ~ . C, ~ H48.1; Z O :H, 6.39; N,9.90; CHaO, 7.29. Found: C, 47.8; H, 5.91; N, 10.1; CHJO, 6.81. d-3-P-Keto-y- (3-hydroxy-2-piperidy1)propyll-4-quinazolone(the Hydrangea alkaloid, febrifugine) (V). A solution of 1.8 g. of IV in 18 cc. of 48% hydrobromic acid was refluxed for ten minutes, then evaporated t o dryness in vacuo. The residue was twice dissolved in 18 cc. of 6 N hydrochloric acid and evaporated t o dryness in vacuo. A solution of the residue in 18 cc. of absolute alcohol was again evaporated t o dryness i n vacuo. The semi-crystalline residue was heated t o boiling with 18 cc. of absolute alcoholic hydrogen chloride for about five minutes. After being cooled to 0" for one hour, the mixture was filtered and the solid washed with two 4-cc. portions of absolute alcoholic hydrogen chloride, then with acetone; yield, 1.05 g. (59%) of white crystals of febrifugine dihydrochloride, m.p. 218-220" dec. with some shrinkage a t 200'. Recrystallization from 10 cc. of 95'70 methanol by the addition of 30 cc. of saturated absolute alcoholic hydrogen chloride raised the m.p. to 232-233' dec., CY]^' $12.8" (0.8% in HzO). This compound had an I.R. spectrum and antimalarial activity identical with the natural Hydrangea alkaloid for which a micro m.p. 223-225' and [CY]:' +12.8" (0.8% in HzO) have been recorded (2). Anal. Calc'd for CleHleNaOa.2HCl: C, 51.3; H, 5.68; N, 11.2. Found: C, 51.5; H, 6.15; N , 11.2. The N-carbamyl derivative, prepared with potassium cyanate (3), formed white crystals from dilute methanol, m.p. 239-241". It gave no depression in m.p. when mixed with the N-carbamyl derivative of the natural alkaloid, m.p. 238-240", and both had identical I.R. spectra.
+
AN ANTIMALARIAL ALKALOID FROM HYDRANGEA. XXI
183
Anal. Calc'd for C I ~ H ~ T C, ~ O59.3; ~ : H, 5.85; N,16.3. Found: C, 59.3;H,6.13; N,16.2. When a sample of the synthetic dihydrochloride was converted t o the high-melting form of the free base, white crystals from chloroform-petroleum ether were obtained, m.p. 158-160", [a]: $16" (0.8% in abs. alc.). Anal. Calc'd for CldIlpNaOr: N, 14.0. Found: N, 14.0. Koepfli, Mead, and Brockman (4) have recorded m.p. 154-156" for the high-melting dimorph, 134-140" for the low-melting dimorph, and the same rotation for both, [a]"," +Bo (0.5% in alc.). Kuehl, Spencer, and Folkers (6)have reported [aID 21" (1.5% in alc.) for the low-melting dimorph. The filtrate from the 1.05 g. of febrifugine dihydrochloride was evaporated to dryness i n vacuo. The residue (500mg.) was dissolved in 10 cc. of water, the solution basified with potassium carbonate, and then extracted with 3 10-cc. portions of chloroform. Dried with magnesium sulfate, the combined extracts were evaporated to dryness i n vacuo (bath 30") leaving 340 mg. of a gum. Crystallization from 3 CC. of acetone was complete after W hour a t room temperature; yield, 175 mg. (13%) of white needles, m.p. 138-139", [a];+121" (0.4% in CHCls). This compound had an I.R. spectrum identical with that of isofebrifugine.' Anal. Calc'd for CldH~pNsOs:C, 63.8;H, 6.40; N, 14.0. Found: C, 63.6;H, 6.63;N, 13.5. The following constants have been recorded for isofebrifugine: m.p. 129-130", [a]: $120" (0.8% in CHCll) (6). $131" (0.4% in CHCI,) (4) and m.p. 131-132",
+
SUMMARY
The syntheses of optically active febrifugine (the Hydrangea alkaloid) and isofebrifugine starting with I-N-benzoyl-p-furyl-@-alaninehave been described. The structures of these alkaloids have been elucidated from their respective infrared spectra. PEARLRIVER,N. Y. REFERENCES (1) BAKER,McEvOY,SCHAUB, JOSEPH, AND WILLIAMS, J . org. Chem., 18,preceding paper
(1953);paper XX of this series. (2) ABLONDI,GORDON, MORTON, AND WILLIAMS, J . Org. Chem., 17, 14 (1952);paper I1 of this series. (3) HUTCHINOS, GORDON, WOLF, ABLONDI,A N D WILLIAMS, J . Org. Chem., 17, 19 (1952); paper I11 of this series. (4) KOEPFLI,MEAD,AND BROCKMAN, J . Am. Chem. Soc., 71, 1048 (1949). (5) KOEPFLI,BBOCKMAN, AND MOFFAT, J . Am. Chem. Soe., 72,3323 (1950). (6) KUEHL,SPENCER, AND FOLKERS, J . A m . Chenz. Soc., 70,2091 (1948). (7) BAKER,SCHAUB, M c E v o ~ AND , WILLIAMS, J . Org. Chem., 17, 132 (1952);paper XI1 of this series. (8) CHOU,Fu,AND KAO,J . Am. Chem. SOC.,70,1765 (1948). (9)BERGMA"AND PINCHAS, Rec. trau. chim., 71, 161 (1952).
